Wikipedia Redefined

I think I can support an idea that would hope to turn otherwise passive consumers into not-so-much-any-longer passive consumers. As long as pseudonyms remain acceptable.

However, this person has some legitimately great ideas. I love how the design is far more reader-centric. I'm not sure why I need a history of articles that I read (browsers do that very well these days), but the 'highlighted' text is a cool idea. You can start thinking about the site as helping you research things, keep a scrapbook of snippets. I love it.

The front page redesign: believe it or not, the multiple languages are the most important thing to highlight. Wikipedia's global audience often uses that system to navigate between encyclopedias. They also often use Google to find the English article, and then look for an 'inter-wiki link' in the margin to an article in their native language.

It looks like there's a lot of cruft in the design, and maybe someone needs to be very bold and piss off a lot of users and force a new interaction pattern. But this stuff is all there for a reason. The 'random article' button is actually one of the most popular features. Really!

As for the proposed branding: first of all, the ideas presented here are not very good. It reminds me of the generic brands at the supermarket. The gossamer rainbow graph wouldn't even reproduce properly at small sizes (and if projects are added or eliminated, then what, do we change the logo?)

But more importantly - the thing which the designers rarely understand is that Wikipedia and its sister projects are not products to be sold - they are communities. And they came to consensus on those logos. They're more like sports team logos than a unified branding system to sell something. That said, there is a system, of sorts; when new logos are made, they try to make variations on the red dot and blue and green shapes.

Also, don't get me started on making color meaningful for navigation. It works for subway maps and it sucks everywhere else. Very bad for accessibility (color-blind people). And very bad for maintainability. The Russian Wikipedia is currently the fastest growing site; you can expect it to change position in the rankings soon. Then what, add another color? Should it change colors, surprising the user? Swap the colors in the rainbow?

Lastly, this designer isn't even addressing the biggest problem we have today, which is how to modify Wikipedia for the mobile web. Reading articles is getting better, and we've been using the Wiki Loves Monuments annual contest as a way to drive the development of mobile photo submissions. But there's still no clear vision of how anyone does serious editing on a mobile device.

As for the part where they offhandedly remark that we should make the site live-editable... HA HA HA. You have no idea what you're up against. I worked on this myself for a while. We made some interesting demos but they weren't something you could deploy.

If we were making Wikipedia from scratch today, of course we'd do that and more, but the thing is, there are multiple challenges, and a whole lot of legacy to support.

Technically: it has to serialize to wikitext and be uploaded as discrete changes to sections. So if you want live editing you need bidirectional parsing and serialization in the browser. Wikitext is unlike any other regular language and has a complex macro system, which consists of... other wiki pages. Stored in the database. Which means you need heavy database I/O just to render HTML. Or at least, a very extensive cache of page fragments. You also can't cheat with a simpler parser in the browser, because wikitext was basically designed to indulge whatever shortcuts the community wanted, and be extremely forgiving. Most wiki pages exploit at least one of the weird quirks. You can't even cheat by regularizing wikitext as you go, because then you're causing spurious changes that the community can't easily police. The current team is solving this with a radical approach to parsing that leverages HTML5's standards and a Node.JS based system. So eventually the parser on the site and in the editor might be very similar.

Operationally: Wikipedia is a cheap site to run because it's basically a static site that you can serve from cache. But changing an article can be monstrously inefficient. There are some articles, like "Barack Obama", that would take minutes to re-render if the caches were empty. When you start changing the basic database model to be more 'live', the costs start to explode.

But rather than drown in negativity, let me just say that whoever this is - thank you for throwing your ideas out there. Assuming this isn't just a resume-building exercise, get in touch with the MediaWiki developers. They need designers.